The Bernstein Center for Computational Neuroscience Freiburg

Announcement for the next

BCCN Seminar

Prof. Carl van Vreeswijk

René Descartes University
Paris
France

Extracting information from Renewal processes

Recently information theory have become more and more prominent in the description of experimental data. However the null hypothesis against which the results are mostly tested is the description of the neuron's activity as Poisson process. The unsurprising result is then that this null hypothesis does not describe the data well. I will discuss an improved null hypothesis, under which the neuronal activity is described by a more general renewal process. This process is still simple enough that information theoretic measures can be calculated under with it. Under such processes the spike count is generally not the best statistic to describe the activity of the neuron, and I propose a better measure. I also show that several features of spike trains, which cannot be captured by Poisson processes are naturally arise in renewal processes and discuss which features of the spike statistics give rise to them.

Host: Prof. Ad Aertsen